[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kfm-devel
Subject:    SMB slave
From:       Jelmer Feenstra <spark () dds ! nl>
Date:       2000-11-16 21:34:45
[Download RAW message or body]

Hi,

I'm sorry I didn't maintain the thread with the discussion about the smb 
slave for konqueror, I just re-subscribed to this mailinglist.

To get a few things straight (I've been closely monitoring samba 
mailinglists, kfm-devel and was in contact with Nicolas Brodu about his lib) :

- The native code present in libsmb++ isn't quite usable (as was said 
before), there are memory leaks and lots of bugs. Updating this code could be 
an option, however I don't think it's an easy thing to do :)

- The link with the samba-tng code works better, however I tried using it 
with some test software I wrote and had some problems with various issues as 
well :( One of the worst is the samba-tng lib not being very up-to-date, it 
can't even list shares of NT5/WIN2K systems. Even if this samba-tng lib 
becomes fully usable it would be better to just rewrite the code of  kio_smb 
(it's not very hard to write read/open-like functions using the samba-tng 
lib) so libsmb++ can be dropped.

- At the moment I had the best results with the code present in the libsmb 
dir in the 'normal' (NOT samba-tng, which has forked as was also metioned :) 
samba source tree, although this is not compiled as a shared library. On the 
samba mailinglist I read a message from a guy that wanted to build a perl 
module to access SMB shares/files, one of the more interesting replies he got 
was the following :

<quote Gerald Carter>
I would stay tuned.  :-)  We are ironically in the process
of experiementing with some shared library support in various
places.  Any I have spoken with a couple other people about
the modularization of code.
</quote Gerald Carter>

So this would mean it would be best (in order to prevent duplicate work being 
done) to make sure those guys from the samba team modularize their code fast 
(which unfortunately isn't very easy either).

- One way not to go (IMHO) would be to make SMB support available by writing 
a wrapper for smbclient because of the 
movie-watching-from-the-network-problem :)

Well, that's about it. I'm afraid we can't do much besides helping the guys 
at the samba project out with building a shared library. I for one know they 
have been talking about it for quite some while now, but I don't think it's 
making a lot of progress :(

Jelmer Feenstra

[prev in list] [next in list] [prev in thread] [next in thread]